Abstract
Ultrasound (US) imaging has been broadly utilized as part of kidney diagnosis because of its ability to show structural abnormalities like cysts, stones, and infections as well as information about kidney function. The main aim of this research is to effectively classify normal and abnormal kidney images through US based on the selection of relevant features. In this study, abnormal kidney images were classified through gray-scale conversion, region-of-interest generation, multi-scale wavelet-based Gabor feature extraction, probabilistic principal component analysis-based feature selection and adaptive artificial neural network technique. The anticipated method is executed in the working platform of MATLAB, and the results were analyzed and contrasted. Results show that the proposed approach had 94% accuracy and 100% specificity. In addition, its false-acceptance rate is 0%, whereas that of existing methods is not <27%. This shows the precise prediction level of the proposed approach, compared with that of existing methods.
IMPLEMENTATION OF ARTIFICIAL NEURAL NETWORK FOR FACE RECOGNITION USING GABOR FEATURE EXTRACTION
